Our research investigations are aimed at developing detailed structural descriptions of reactive molecule and relating these observations to their biological funciton. These investigations fall under the broader class of problems involving molecular recognition, and are grounded in the view that a systematic structural characterization of molecular interactions can be used to a) relate molecular structure to function, b) identify new modecular targets available for therapeutic intervention, and c) discover powerful therapeutic candidates that interact with these targets. Our approaches are based on systematic observations of large informational databases containing descriptions of molecular structures and molecular reactions. Analysis of this information is used to develop models with low, intermediate and high degrees of structural detail. Using these models, three areas of study have been completed. First, target molecules are thought to exhibit flexibility and in many instances this flexibility is directly involved in biological function. Low resolution models of protein backbones have been used to explore structural features and relate them to molecular flexibility. Second, close atomic contacts are known to be essential for ligand binding and the formation of stable folded protein molecules. A variety of intermediate resolution models have been explored, based primarily on considerations of atomic packing. These models characterize ligand binding surfaces in terms of shape and chemical features and use this information to identify receptor targets on molecular surfaces and structurally important regions on the interiors of folded proteins. Third, many bioreactive ligands act by binding to their target's surface and then reacting with key atoms. High resolution models of detailed quantum chemical descriptors have been used to characterize the electronic structure of reactive partners. These investigations have led to the development of strategies for the design of ligands that are reactive against classes of proteins that contain zinc-finger domains. Nearly 2 percent of the proteins in the human genome are thought to contain zinc fingers, with many known to be important in transcriptional regulation. Efforts to identify novel ligand against selected zinc-finger containing proteins have lead to a new class of chemotypes active against retroviral nucleocapsid zinc-fingers.Our current and future research efforts aim at the development of a greater understanding of the atomic details of molecular interactions. This information will be used to develop novel computational tools that can be directly applied to the process of new drug discovery as it relates to anticancer agents. This effort will utilize the available databases obtained from synthetic and natural products that have been tested in the National Cancer Institutes cell screening projects. The long-term goal is to investigate relationships between unique patterns of cell screening test results with patterns of structural similarity within tested and untested compounds.Z01-BC-10281